| name | stats-tracker |
| description | Track Claude Code usage statistics and skill effectiveness for CircleTel. Monitor productivity, model usage, streaks, and skill performance metrics. |
| version | 2.0.0 |
| dependencies | error-registry, compound-learnings |
Stats Tracker
Skill for tracking Claude Code usage statistics AND skill effectiveness (RSI metrics).
When to Use
This skill activates when you:
- Want to see your usage statistics
- Track productivity over time
- Analyze model usage patterns
- Monitor usage streaks
- Plan based on usage limits
- NEW: Review skill effectiveness metrics
- NEW: Generate weekly effectiveness reports
Keywords: stats, usage, analytics, productivity, streak, model usage, tokens, sessions, cost, skill effectiveness, skill metrics
Quick Commands
| Command | Description |
|---|
/stats | View usage stats, streak, favorite model |
/stats skills | View skill effectiveness dashboard |
/stats skills [name] | Deep-dive on specific skill |
/stats verification | NEW: View pass@k metrics by verification type |
/stats verification log | NEW: Log a verification session result |
/usage | View token usage and limits |
/context | View current context usage |
/cost | View session cost breakdown |
Understanding /stats Output
/stats
๐ Your Claude Code Stats
โโโโโโโโโโโโโโโโโโโโโโโโโโโโ
๐ Favorite Model: Sonnet 4 (78% of sessions)
๐ฅ Current Streak: 12 days
๐ This Week: 847K tokens across 23 sessions
๐ Usage Graph: [โโโ
โโโโโ] (last 7 days)
Metrics Explained
| Metric | Description |
|---|
| Favorite Model | Most frequently used model |
| Current Streak | Consecutive days using Claude Code |
| This Week | Total tokens and sessions |
| Usage Graph | Visual of daily usage pattern |
Understanding /usage Output
/usage
๐ Usage This Period
โโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Sonnet: โโโโโโโโโโ 78%
Opus: โโโโโโโโโโ 15%
Haiku: โโโโโโโโโโ 7%
Resets: Monday 00:00 UTC
Model Selection Guide
Choose the right model for the task:
| Task Type | Model | Why |
|---|
| General coding | Sonnet | Best balance of speed/quality |
| Complex architecture | Opus | Deep reasoning, long context |
| Quick fixes | Haiku | Fast, efficient |
| Code exploration | Haiku (Explore) | Optimized for search |
| Documentation | Sonnet | Good writing quality |
| Bug investigation | Sonnet/Opus | Depends on complexity |
Cost Efficiency
| Model | Relative Cost | Best For |
|---|
| Haiku | $ | Quick tasks, exploration |
| Sonnet | $$ | Most development work |
| Opus | $$$ | Complex problems |
Daily Workflow with Stats
Morning Routine
1. /stats # Check streak, plan day
2. /usage # Check remaining budget
3. /rename feature-x # Name today's session
4. Start coding # Use appropriate model
During Development
# After complex task
/context # Check context usage
# When switching tasks
/stats # Quick progress check
End of Day
/stats # Review productivity
/usage # Check usage remaining
/cost # See session cost
Usage Optimization Strategies
If Usage is High
- Use Haiku more - For exploration and quick tasks
- Spawn Explore agents - Uses efficient Haiku model
- Background tasks - Don't consume extra tokens
- Be more targeted - Specific queries use less context
If Starting Fresh Week
- Plan major tasks - Allocate Opus for complex work
- Default to Sonnet - Best general-purpose model
- Reserve Haiku - For quick lookups and exploration
Maximizing Efficiency
# Good: Specific query
"Fix the null check on line 45 of auth.ts"
# Bad: Vague query (uses more context/tokens)
"Help me with the auth system"
Tracking by Feature
Use named sessions to track usage per feature:
/rename billing-feature
/stats
"How many tokens did billing vs auth use?"
Weekly Review Pattern
Every Friday:
1. /stats # Review week's usage
2. /usage # Check budget status
3. Note completions # What features finished?
4. Plan next week # Allocate model usage
Weekly Goals Template
Week Goals:
- [ ] Maintain 5-day streak
- [ ] Complete 3 features
- [ ] Use Opus only for architecture decisions
- [ ] Try Explore agent for research tasks
- [ ] Stay under 80% usage
Integration with Other Skills
With Session Manager
# Track stats per named session
/rename billing-feature
[work on feature]
/stats
# See usage specific to this session
With Context Manager
# Monitor both context and usage
/context # Current context budget
/stats # Overall usage patterns
With Async Runner
# Background tasks don't inflate usage
"Run build in background"
# Build output doesn't use tokens
Gamification Ideas
Streak Challenges
- 5-day streak: Bronze
- 10-day streak: Silver
- 30-day streak: Gold
- Try to never break your streak!
Model Efficiency Challenge
Challenge: Complete task with Haiku
1. Try Haiku first for any task
2. Escalate to Sonnet only if needed
3. Use Opus sparingly for max impact
Weekly Competition (Team)
- Compare completed features
- Compare usage efficiency
- Share tips for optimization
External Tools
ccusage (CLI Analytics)
npm install -g ccusage
ccusage
ccusage --daily
ccusage --monthly
Claude Code Usage Monitor
Real-time terminal monitoring with:
- Token tracking
- Burn rate analysis
- Predictions for limits
CircleTel Productivity Patterns
Feature Development Session
Start: /rename feature-customer-billing
/stats # Baseline
Middle: [coding work]
/context # Check context
End: /stats # Compare to baseline
/cost # Session cost
Bug Fix Session
/rename BUG-1234-fix
/stats # Note starting point
[investigation and fix]
/stats # See investigation cost
# Typically: Sonnet for analysis, quick fixes
Architecture Planning
/rename architecture-review
# Use Opus for deep thinking
"Think about the best approach for..."
/stats # Higher usage expected
/cost # Worth it for good architecture
Skill Effectiveness Tracking (RSI Metrics)
NEW in v2.0: Track which skills perform best and feed insights back for improvement.
Command: /stats skills
/stats skills
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
SKILL EFFECTIVENESS DASHBOARD - February 2026
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
MOST EFFECTIVE (by success rate) โ NEEDS IMPROVEMENT
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโผโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
1. database-migration 95% โ โ 1. coverage-check 70% โ
2. bug-fixing 89% โ โ 2. refactor 75% โ
3. compound-learnings 87% โ โ
โ
MOST USED (activations) โ UNDERUTILIZED
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโผโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
1. bug-fixing 47 times โ 1. mobile-testing 2 times
2. context-manager 35 times โ 2. deployment-check 3 times
3. stats-tracker 28 times โ
INSIGHT: bug-fixing success rate improved 12% after adding
error-registry integration last week.
RECOMMENDATION: Consider promoting mobile-testing skill
(high success rate, low activation).
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Metrics Tracked Per Skill
| Metric | Description | How Measured |
|---|
| Activation Count | Times skill triggered | Keyword detection |
| Success Rate | Completed without correction | No follow-up corrections |
| Resolution Time | From activation to completion | Session timestamps |
| Correction Rate | How often skill was corrected | Links to compound-learnings |
| Pattern Contribution | New patterns generated | Links to learnings/ |
Deep-Dive: /stats skills [name]
/stats skills bug-fixing
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
SKILL DEEP-DIVE: bug-fixing
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Version: 1.1.0
Dependencies: error-registry
METRICS (February 2026)
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Activations: 47
Success Rate: 89%
Avg Resolution: 23 min
Corrections: 7 (15%)
Patterns Created: 3
PHASE BREAKDOWN
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Phase 0 (Registry): Avg 2min โ Skip rate: 20%
Phase 1 (Understand): Avg 5min โ Skip rate: 10%
Phase 2 (Investigate): Avg 10min โ Skip rate: 5%
Phase 3 (Fix): Avg 5min โ Success: 95%
Phase 4 (Validate): Avg 3min โ Skip rate: 30%
TREND
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Success rate: โ 12% since error-registry integration
Registry hits: 40% of bugs found in known patterns
RECENT CORRECTIONS (to improve)
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
- 2026-02-10: Missed RLS policy check
- 2026-02-08: Used wrong column name
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Data Storage
Skill metrics are stored in:
.claude/skills/stats-tracker/
โโโ metrics.json # Current period metrics
โโโ verification-metrics.json # Pass@k verification metrics
โโโ reports/
โโโ weekly-YYYY-WW.md # Weekly reports
Verification Metrics (Pass@k Tracking)
Track how many verification attempts are needed before success.
Command: /stats verification
/stats verification
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
VERIFICATION METRICS - March 2026
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
FIRST-ATTEMPT SUCCESS (pass@1)
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Type-check: โโโโโโโโโโ 84% (target: 90%)
Build: โโโโโโโโโโ 78% (target: 85%)
Unit tests: โโโโโโโโโโ 72% (target: 75%) โ
E2E: โโโโโโโโโโ 58% (target: 60%)
Lint: โโโโโโโโโโ 96% (target: 95%) โ
TREND (pass@1 rate, last 4 weeks)
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Week 10: โ 78%
Week 11: โ
82%
Week 12: โ 85% โ
COMMON FIX CATEGORIES
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
1. Code changes: 62%
2. Config tweaks: 18%
3. Missing imports: 12%
4. Test adjustments: 8%
INSIGHT: Type-check pass@1 improved 7% after adding
type-guards-optionals.md rule.
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Understanding Pass@k
| Metric | Meaning | Target |
|---|
| pass@1 | Passed on first attempt | > 80% |
| pass@3 | Passed within 3 attempts | > 95% |
| pass@5 | Passed within 5 attempts | 100% |
| fail@5+ | Needed 5+ attempts | < 2% |
Grader-Specific Targets
| Verification Type | pass@1 Target | Rationale |
|---|
| type-check | 90% | Type errors are predictable |
| build | 85% | Env/import issues common |
| unit-test | 75% | Logic bugs take iteration |
| e2e | 60% | Integration complexity |
| lint | 95% | Mostly auto-fixable |
| manual | 80% | Subjective criteria |
Logging a Verification Session
After completing verification, log the result:
/stats verification log type-check 2 code
# Logs: type-check verification, passed on attempt 2, fix was code-related
Format: /stats verification log <type> <attempts> <fix_category>
Fix Categories
| Category | Examples |
|---|
code | Wrong type, missing property, logic error |
config | tsconfig, package.json, env vars |
dependency | Missing import, version mismatch |
test | Test assertion wrong, fixture issue |
RSI Integration
Low pass@1 rates trigger investigation:
pass@1 < target
โ
โผ
Analyze fix_categories
โ
โผ
Common pattern? โโโโบ Create compound-learning
โ
โผ
Extract rule โโโโโโโโโบ Add to .claude/rules/
โ
โผ
pass@1 rate improves
Weekly Verification Review
Add to your Friday review:
/stats verification โ Check pass@1 rates- Compare to targets โ Which types need improvement?
- Review fix categories โ What's causing failures?
- Consider new rules โ Can a pattern prevent failures?
Weekly Report Generation
Every Monday, generate a skill effectiveness report:
# Skill Effectiveness Report - Week 7, 2026
## Summary
- Total skill activations: 142
- Average success rate: 85%
- Most improved: bug-fixing (+12%)
- Needs attention: coverage-check (70%)
## RSI Loop Impact
- Error registry hits: 40% (saving ~10min per known bug)
- Corrections captured: 5
- Rules extracted: 2
- Patterns created: 3
## Recommendations
1. Add more MTN API patterns to error-registry
2. Update coverage-check with fallback providers
3. Promote mobile-testing (underutilized, high success)
Integration with RSI Skills
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ RSI METRICS FLOW โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโค
โ โ
โ SKILL ACTIVATION โโโบ TRACK METRICS โโโบ WEEKLY ANALYSIS โ
โ โ โ โ
โ โ โผ โ
โ โ GENERATE INSIGHTS โ
โ โ โ โ
โ โ โโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโ โ
โ โ โ โ โ
โ โผ โผ โผ โ
โ CORRECTION? โโโบ compound-learnings SKILL UPDATE โ
โ โ โ โ
โ โผ โ โ
โ error-registry โโโโโโโโโโโโโโโโโโโโโโโโโ โ
โ โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Best Practices
- Check stats daily - Start with
/stats
- Match model to task - Don't use Opus for simple fixes
- Track by feature - Use named sessions
- Review weekly - Adjust workflow based on patterns
- Use Haiku for exploration - Spawn Explore agents
- Background for builds - Don't waste tokens on output
- Targeted queries - Specific questions = efficient usage
- NEW: Review
/stats skills weekly for improvement opportunities
Troubleshooting
Stats seem wrong
# Check current session specifically
/cost
# Check usage for period
/usage
High usage unexpectedly
- Check for large file reads
- Review if background tasks are actually background
- Consider if queries could be more specific
Streak reset
- Streak requires at least one interaction per day
- Even
/stats counts
- Check timezone (UTC-based)
Version: 2.0.0
Last Updated: 2026-02-12
For: Claude Code v2.0.64+
RSI Integration: error-registry, compound-learnings
